On November 13 Google introduced Gemini 4 Hydrogen — a model positioned as a fast workhorse for agentic scenarios: research, data analysis, coding, testing and codebase navigation.

Pricing and specs

  • $1.50 per million input tokens, $7.50 per million output tokens.
  • 1 million token context window.
  • Claimed: tasks run 2.5x faster than Gemini 3.8 Flash.
  • Four reasoning levels — save tokens on simple steps.

Antigravity and the orchestrator

Antigravity gained an Orchestrator that switches models mid-session: cheap for routine edits, expensive for hard decisions. Google reports near-Carbon task success with 24% less quota and 19% faster completion.

The caveat

Google warns plainly: parallel agent teams burn quota faster. Before delegating a whole project, compare a small measurable change against your normal workflow. Vendor benchmarks do not prove your real costs.